impeccable

Designs, audits, and refines production-grade frontend interfaces across web and native platforms.

Updated Jul 7, 2026
One-click install
npx skills add https://github.com/singhaganesh/Urban-assist --skill impeccable-singhaganesh
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: impeccable
Source: https://github.com/singhaganesh/Urban-assist/tree/main/.cursor/skills/impeccable
Command: npx skills add https://github.com/singhaganesh/Urban-assist --skill impeccable-singhaganesh

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? AI-generated interfaces tend to converge on the same generic patterns: cream backgrounds, gradient text, identical card grids, and eyebrow labels on every section. This Skill provides a structured design workflow that produces distinctive, production-grade frontend code while systematically eliminating recognizable AI design tells. ## Core Features & Use Cases - Command-driven design workflow: 23 sub-commands organized into Build (craft, shape, init, document, extract), Evaluate (critique, audit), Refine (polish, bolder, quieter, distill, harden, onboard), Enhance (animate, colorize, typeset, layout, delight, overdrive), Fix (clarify, adapt, optimize), and Iterate (live) categories. - Register-aware guidance: Separate references for brand surfaces (landing pages, marketing) versus product UI (dashboards, apps), plus platform-specific conventions for iOS (HIG) and Android (Material 3). - Context detection scripts: Node scripts detect project context (PRODUCT.md, DESIGN.md), generate brand palette seeds in OKLCH, and scan markup for design anti-patterns. - Use Case: A developer asks to audit a Next.js dashboard; the Skill loads the product register reference, scores accessibility, performance, theming, responsive design, and anti-patterns on a 0-20 scale, then recommends prioritized follow-up commands. ## Quick Start Ask the assistant to use the impeccable skill to critique the design of your landing page or run /impeccable audit on a specific UI surface.

Frequently Asked Questions about impeccable

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I audit a web UI for accessibility and performance issues?

Invoke the audit command on a target surface. It scores five dimensions (accessibility, performance, theming, responsive design, anti-patterns) from 0-4 each, reports findings by P0-P3 severity, and recommends follow-up commands like polish or adapt.

What commands does the impeccable design skill support?

It provides 23 commands grouped by intent: craft, shape, init, document, and extract for building; critique and audit for evaluation; polish, bolder, quieter, distill, harden, and onboard for refinement; animate, colorize, typeset, layout, delight, and overdrive for enhancement; clarify, adapt, and optimize for fixes; plus live for browser-based iteration.

Does the skill support native iOS and Android app design?

Yes. When the project platform is ios, android, or adaptive, it loads platform-specific references covering Apple HIG conventions and Material Design 3, with native variants of the audit and adapt commands. Web-only tooling like the live browser overlay is disabled for native projects.

How does the skill avoid generic AI-generated design patterns?

It enforces absolute bans on common AI tells such as gradient text, glassmorphism, hero-metric templates, side-stripe borders, and repeated eyebrow labels. It also applies a two-level slop test checking whether the theme and aesthetic family are predictable from the product category.

When should I use craft versus shape versus polish?

Use shape to plan UX before coding and craft to plan then build a feature end-to-end. Use polish as the final quality pass on existing work before shipping. Most refine and enhance commands hand off to polish as the last step.

What are the limitations of the live browser iteration mode?

The live command and the bundled detector are web-only; they do not apply to ios, android, or adaptive native projects. Live mode also requires a running dev server, and the detector reads local HTML and CSS files without network access.